Cerebrospinal fluid (CSF) boosts metabolism and virulence expression factors in Acinetobacter baumannii

In a recent report by the Centers for Disease Control and Prevention (CDC), multidrug resistant (MDR) Acinetobacter baumannii is a pathogen described as an “urgent threat”. Infection with this bacterium manifests as different diseases such as community and nosocomial pneumonia, bloodstream infections, endocarditis, urinary tract, wound infections, burn infections, skin and soft tissue infections, and meningitis. In particular, nosocomial meningitis, a common complication of neurosurgery caused by extensively-drug resistant (XDR) A. baumannii, is extremely challenging to manage. Therefore, it is necessary to identify signals, such as exposure to cerebrospinal fluid (CSF), that trigger expression of virulence factors that are associated with the successful establishment and progress of this infection. While a hypervirulent A. baumannii strain did not show changes in its transcriptome when incubated in the presence of CSF, a low-virulence isolate showed significant differences in gene expression and phenotypic traits. Exposure to 4% CSF caused increased expression of virulence factors such as fimbriae, pilins, and iron chelators, and virulence as determined in various model systems. Furthermore, although CSF’s presence did not enhance bacterial growth, it was associated with an increase of expression of genes encoding transcription, translation, and the ATP synthesis machinery. Experiments to identify the active CSF component pointed to human serum albumin (HSA). Importance Acinetobacter baumannii, notorious for its multidrug resistant phenotype, overcomes nutrient deprived and desiccated conditions through its metabolic flexibility, pathogenic and physiological adaptability. Although this pathogen is commonly associated with respiratory infections, there have been a considerable amount of cases of A. baumannii bacterial meningitis. These infections are usually post-neurological surgery complications associated with high mortality rates ranging from 40 to 70%. This work describes interactions that may occur during A. baumannii infection of human cerebrospinal fluid (CSF). A. baumannii’s displays capabilities to persist and thrive in a nutrient-limited environment, which also triggers the expression of virulence factors. This work also further explores A. baumannii’s utilization of an essential component within CSF to trigger enhanced expression of genes associated with its pathoadaptibility in this environment.

Acinetobacter baumannii has emerged as an important pathogen due to its ability to 74 resist multiple antibiotics, persist in hospital settings, and cause a wide variety of 75 infections such as pneumonia, bacteremia, urinary tract infections, skin and soft-tissue 76 infections, and meningitis (1-4). The acquisition of resistance to carbapenems by 77 certain strains (carbapenem-resistant Acinetobacter baumannii, CRAB) increased the 78 problematic nature of this pathogen (5), which has been qualified as an "urgent" threat 79 in a recent report by the Centers for Disease Control and Prevention (6). 80 Bacterial meningitis, which is considered a medical emergency, is a serious 81 infection that can cause permanent disabilities (brain damage, hearing loss, and 82 learning problems) or death if untreated (7-9). Post neurosurgical A. baumannii 83 meningitis can cause death or leave permanent sequelae and is usually associated with 84 high mortality rates reaching up to 40 to 70 %. (10,11). Illustrating the dangerous 85 nature of these infections is the recent case of the A. baumannii infection of a 39-year-86 old man treated with external ventricular drainage of cerebrospinal fluid (CSF). Although 87 the strain was susceptible to colistin at the time of detection, it quickly acquired 88 resistance without losing virulence (12). This genetic plasticity, a consequence of its 89 ability to acquire and integrate foreign DNA, gives A. baumannii's a tremendous 90 metabolic versatility that permits the bacterium to adapt and persist in harsh conditions 91 (2,(13)(14)(15)(16)(17). A. baumannii's success in causing numerous infections, where it gets in 92 contact with different body components and fluids, must be the result of its capabilities 93 to not only capture adequate genetic determinants but also regulate expression of the 94 proper cell components (2,(13)(14)(15)(18)(19)(20). We have also previously documented that 95 baumannii's response to different stimuli is dependent on each particular strain's degree 119 of pathogenicity. Less pathogenic strains induced more changes in their phenotypic 120 behavior to overcome the stressful environment and persist (23). 121 The analysis of A. baumannii A118 DEGs revealed an increase in the expression 122 of many genes involved in the manifestation of genetic information and energy 123 production machineries (Table S1). Notably, a large proportion of ribosomal protein 124 genes are overexpressed upon exposure to CSF. Among the ribosomal protein 125 associated genes, 47 out of 55 displayed a significant increase of expression of 2-fold 126 or more. Coincidently, key translation genes such as those encoding elongation factors 127 (EF) EF-G, EF-F and EF-P were also up-regulated. Concurrently, the main genes of the 128 transcriptional machinery (RNA polymerase) were similarly overexpressed. The rpoB 129 and rpoC genes, which code for the beta and beta' subunits of RNA polymerase (core of 130 the transcription machinery), were overexpressed with a log2fold just below 1. However, 131 the gene encoding the alpha subunit was also up-regulated with a log 2 fold change of 132 1.48 (Table S1). In addition, genes important for energy production in the cell were also 133 induced upon CSF exposure. The atpIBEFHAGCD locus, an operon encoding the FoF 1 -134 ATP synthase (the main ATP generator in the bacterial cell) displayed a 3-fold 135 transcriptional increase in expression (Table S1). These transcriptional responses 136 suggest that A. baumannii responds to CSF by overexpressing machinery involved in 137 gene expression (transcription and translation genes) and the main ATP generator, 138 FoF1-ATP synthase. Also, CSF exposure induces the transcription of specific metabolic 139 routes in A. baumannii. In particular, several dehydrogenases of tricarboxylic acid cycle 140 (TCA) intermediates as well as a citrate symporter are significantly overexpressed, 141 together with two proline symporters (Table S1). 142 Gene ontology (GO) was performed to identify molecular functions and biological 143 pathways associated to A. baumannii's adaptive responses to CSF. Consistently with 144 the above mentioned DEGs, GO enrichment analysis revealed a statistically significant 145 overrepresentation of the GO categories ATP synthesis coupled proton transport, 146 translation, tricarboxylic acid cycle, and aerobic respiration by 13. 8-, 8.8-, 4.9-and 4.3-147 fold, respectively (P-value < 0.05). Other authors found that exposure of A. baumannii 148 strains to amikacin, imipenem, and meropenem was associated with an increase in the 149 expression of genes involved in the TCA cycle, biosynthesis of amino acids, purines, 150 and pyrimidines, as well as, the operons involved with ATP, RNA, and protein synthesis 151 (24). Taken together, these results suggest that various stressful environments 152 (nutrients availability and antibiotic treatment) induce expression of genes related to 153 energy production, protein synthesis, and metabolism in A. baumannii. As a 154 consequence of these transcriptomic changes, the bacterial cell must adapt to be able 155 to survive and even thrive in these hostile conditions. 156 CSF boosts specific metabolic routes without increasing growth in optimal 157 growth conditions. 158 The CSF-mediated upregulation of genes coding for the elements necessary for 159 transcription, translation, expression and ATP synthesis was not accompanied by a 160 decrease in generation time ( Figure 1). These results suggest that cells respond to the 161 presence of CSF by enhancing the expression of pathways that produce specific effects 162 rather than increasing growth capacity. To better understand this behavior, we 163 assessed the effect of CSF on cells growing in the presence of different carbon sources. 164 A. baumannii A118 and AB5075 were cultured in minimal medium containing proline, 165 glutamine, glucose, or citrate, all of which are utilized through the tricarboxylic acid cycle 166 and whose expression was modified in the presence of CSF. Addition of CSF to a 167 glutamine-containing minimal medium resulted in no change (A. baumannii A118) or a 168 reduction (A. baumannii AB5075) of the growth rate ( Figure S1). On the other hand, 169 when the carbon source was proline, both strains grew at a higher rate ( Figure S1). The 170 transcriptomic data showed that type I glutamine synthetase (AbA118F_3228) and the 171 proline symporter putP genes are upregulated by a log 2 fold change of 0.58 and 0.87 172 respectively. Studies on Salmonella typhimurium showed that putP codes for a proline 173 permease, an integral membrane protein, that is the primary transport protein when this 174 amino acid is the only carbon or nitrogen source (25). The CSF-mediated increase in 175 abundance of the proline permease in A. baumannii may explain the increase in growth 176

rate. 177
CSF also induced an increase in the growth rate of both strains, although more 178 pronounced in the case of A. baumannii AB5075, when cultured using citrate as sole 179 carbon source ( Figure S2). The transcriptomic analysis showed upregulation (log2fold 180 change 0.62) of AbA118F_2664, a gene that encodes a CitMHS citrate-magnesium 181 hydrogen complex symporter (Table S1). Proteins of the CitMHS family transport 182 citrate-Mg 2+ complex symport with one proton per complex molecule (26). It is of interest 183 that increased citrate levels help survival of A. baumannii in certain conditions (22).  There are numerous reports supporting the hypothesis that increasing the 204 expression of enzymes involved in transcription, translation, and synthesis of ATP is 205 correlated with an increase in bacterial growth rate (28-32). However, these growth 206 differences were not evident in either of the A. baumannii strains in the presence of 207 CSF. An attractive hypothesis to explain this observation is that the increase in gene 208 expression capabilities is channeled toward the synthesis of cell components necessary 209 for survival in the human body, e.g., adhesins and pilins. 210 The data described in this section indicates that certain modifications in the A. of metabolism instead of maximizing growth. In another study that tested the 218 relationship of gene expression with metabolism and growth, it was observed that when 219 bacterial cells were cultured in poor nutrient medium, there was a higher expression of 220 metabolic proteins, such as enzymes and transporters (27). 221 Our data also suggests that in depleted medium such as CSF, A. baumannii may 222 be allocating all possible resources towards metabolism using an uncoupled 223 metabolism to optimize its survival. 224

CSF affects the expression of A. baumannii virulence genes 225
Addition of CSF to A. baumannii A118 cultures induces an increase in the expression of 226 a set of genes that code for virulence-associated functions such as type IV pili, iron 227 uptake systems, the type VI secretion system (T6SS), and poly-N-acetylglucosamine 228 (PNAG) production. 229 Type IV pili participate in microbial adherence as well as motility (gliding or 230 twitching). While A. baumannii lacks flagellum-mediated motility, twitching, and surface-231 associated motility was demonstrated in several strains (28,29). Numerous studies on 232 twitching and surface-associated motility in A. baumannii A118 showed dependence on 233 changes in light and temperature (30) as well as on the components of the growth 234 media. In particular, addition of HSA resulted in increased motility and concomitantly 235 upregulation of the cognate genes (22). Exposure of A. baumannii A118 to CSF 236 increased the expression of pilW (log2fold 1.22), pilJ (log2fold change 0.43), fimA 237 (log2fold change 3.09), fimB (log2fold change 2.32), and the fimbrial protein precursor 238 AbA118F_3133 (log2fold change 1.91). All of the type IV fimbriae genes have been 239 experimentally shown to be associated with motility, cell adhesion, and biofilm formation 240 The presence of CSF was also correlated with higher expression of twelve genes 242 associated with the acinetobactin iron uptake system ( Figure 2 and Table S1). These 243 genes are part of the ferric-acinetobactin receptor-translocation machinery (bauABDE, 244 bauA log 2 fold change 1.67), the acinetobactin biosynthesis (basBDFGJ, basD log 2 fold 245 change 1.86) and export (barB) (33) (34) (Figure 2A and Table S1). Besides their direct 246 role in iron uptake in the iron starvation conditions found in the human host, the 247 products of basD and basA are needed for A. baumannii to persist and cause apoptosis 248 of human alveolar epithelial cells (35). Bacterial iron uptake systems that are virulence 249 factors are usually highly regulated and are induced in conditions of iron starvation. It is 250 was of interest that in A. baumannii A118, genes that code for functions in siderophore 251 biosynthesis, export, and import are upregulated of in the presence of CSF. This finding 252 adds another regulatory signal that enhances expression of acinetobactin iron uptake 253 system. This increase in expression could be directly related to growth in the host or to 254 biofilm formation, which is dependent on efficient iron uptake (36). 255 The T6SS of many bacteria evolved to become essential for virulence (37). They 256 are composed of 14 genes that code for the three components, the phage tail-like, the 257 base plate, and the membrane complexes. Nine out of the 14 A. baumannii A118 T6SS 258 genes were slightly but significantly upregulated in the presence of CSF. These genes 259 tssABCDHIKLM, code for essential components of the T6SS (Figure 2A and Table S1). 260 The structures of bacterial biofilms are usually dependent on polysaccharides 261 such as poly-β-1,6-N-acetyl-d-glucosamine (PGA) or cellulose. Previous studies 262 showed that functional production of PGA in Escherichia coli depends on the products 263 of four genes, pgaABCD. pgaC and pgaD are essential for biosynthesis, and pgaB, 264 which specifies a N-deacetylase, together with pgaA are needed to export the 265 polysaccharide from the periplasm to the extracellular milieu (38). All four homologs 266 were significantly upregulated when A. baumannii A118 was cultured in the presence of 267 CSF (log2fold of 1.63, 1.63, 1.56 and 1.08 for pgaA, pgaB, pgaC and pgaD, 268 respectively) ( Figure 2A and Table S1). As expected, Congo red staining showed that 269 A. baumannii A118 cells cultured in the presence of CSF produced higher levels of PGA 270 ( Figure 2B). 271

CSF enhances the release of A. baumannii's cytotoxic agents. 273
An initial assessment of the effect of CSF on A. baumannii virulence was determined 274 using cytotoxicity assays. Cell-free conditioned medium (CFCM) obtained from A. 275 baumannii A118 and AB5075 cultured in LB with or without CSF was added to human 276 embryonic kidney cells (HEK-293), and the cells were inspected after 1 hour. Figure 2C  277 shows that CFCM samples obtained from CSF-containing A. baumannii AB5075 278 cultures were significantly more cytotoxic than CFCM from cultures that lacked CSF. 279 This increase in cytotoxicity was observed at all tested concentrations, 1.5% CFCM (P-280 value = 0.006), 5% CFCM (P-value = 0.001), and 50% CFCM (P-value < 0.0001). 281 Conversely, CFCM obtained from A. baumannii A118 cultures containing CSF showed 282 an increased cytotoxic effect only at the highest concentration tested (50%) (P-value = 283 0.002). Although at different levels, the results of these assays suggest that the 284 presence of CSF induces the release of one or more cytotoxic substances by A. 285 baumannii ( Figure 2C). 286

CSF-treatment increases A. baumannii virulence. 288
To effect of CSF on A. baumannii's virulence was tested using two models of infection. baumannii A118 or AB5075 cultured in LB medium containing 4% CSF, and inspected 292 at five and ten days post infection. Both strains showed no significant increase in killing 293 after five days but A. baumannii AB5075 did show a significant increase in killing after 294 ten days. Survival with respect to individuals inoculated with cells cultured in LB being 295 was reduced by 3.15% (P-value=0.186) ( Figure 2D). This study was done using a highly 296 robust D. melanogaster population previously shown to have higher stress resistance 297 than commonly used inbred fly lines (40). 298 Infection assays using the Galleria mellonella model showed a statistically 299 significant difference in the killing between A. baumannii A118 cultured in LB or LB plus 300 4% CSF. As in the D. melanogaster model, the bacteria cells cultured in the presence 301 of CSF were more virulent ( Figure 2E). These results showed a correlation between the 302 transcriptional changes in virulence gene expression observed in vitro with an increase 303 in virulence as determined by tests using two different models of infection. 304 HSA plays a role in A. baumannii pathoadaptation when exposed to CSF 305 Previous work showed that the presence of PF is correlated with modifications of the 306 expression of more than 1100 A. baumannii genes including many virulence factors 307 such as motility, biofilm formation, efflux, T6SS, fibrinolytic activity and capsule genes 308 To test this hypothesis, A. baumannii A118 cells were cultured in LB or LB 317 supplemented with one of the following: CSF, HSA-depleted CSF (dCSF), or HSA. Total 318 RNA was extracted from cells growing in all four conditions, retrotranscribed, and the 319 cDNA was used as template for quantitative polymerase chain reactions (RT-qPCR). 320 We assessed levels of expression of the Type 1 fimbrial protein (fimA), rpmC, and atpB 321 (ATP synthase beta (AbA118F_0480), which were among the most DEGs when CSF 322 was present in the culture broth. The addition of CSF to LB increased transcription 323 levels while addition of dCSF resulted in a reduction in levels of expression of fimA and 324 rpmC while little to no changes in expression were observed for atpB (Fig 3A, B and C  325 respectively). The enhancing effect observed in the presence CSF was even more 326 pronounced when the cultures took place in LB containing HSA. In this case, fimA, 327 atpB, and AbA118F_2933 expression levels were 8-fold, 109-fold, and 6-fold higher, 328 respectively ( Figure 3A-C). 329 The expression levels of FoF1-ATP synthase, as determined by ATP synthesis in 330 cells cultured in LB containing CSF or LB containing has, were significantly higher than 331 those observed when the growth media was LB or LB supplemented with dCSF ( Figure  332 3D). These results are in agreement with those obtained by transcriptomics and RT 333 PCR (Table S1, Figure 3C). 334  virulence and is resistant to many antibiotics (51). 385

RNA extraction and sequencing 386
A. baumannii colonies (A118 and AB5075) were suspended in Luria Bertani broth (LB) 387 with or without 4% CSF and incubated with agitation for 18 h at 37°C. Overnight 388 cultures were then diluted 1:10 in fresh LB broth and incubated with agitation for 7 h at 389 37°C. RNA was extracted from each strain using the TRI REAGENT® Kit (Molecular 390 Research Center, Inc., Cincinnati, Ohio, USA) as previously described (22). Total RNA 391 extractions were performed in two biological replicates for each condition. 392 RNA sequencing was outsourced to Novogene (Novogene Corporation, CA) for mRNA-393 seq analysis, which includes rRNA depletion, library preparation through Next® Ultra™ 394

RNA-seq Data Analysis. 397
RNA-seq reads (GEO accession No GSE153967) corresponding to A. baumannii strain 398 A118 and AB5075 exposed to LB or LB plus 4% CSF were analyzed as follows. 399 Trimming of low-quality bases at the ends of the reads to a minimum length of 100 bp 400 and removal of Illumina adaptor sequences was performed using Trimmomatic (52). iGO terms were retrieved from UniProt for the best BLASTx hits to A. baumannii A118F 413 genes. Using GO.db Bioconductor annotation data package in R language, GO terms 414 and ancestor terms were assigned for all DEGs from this study. GO enrichment analysis 415 was performed using custom-made scripts as described previously (55). The 416 enrichment factor was estimated as the ratio between the proportions of genes 417 associated with a particular GO category present in the dataset under analysis, relative 418 to the proportion of the number of genes in this category in the whole genome. p-values 419 were calculated using the Fisher Exact Test and adjusted by the Benjamini-Hochberg 420 method. 421

Growth Curves. 422
Both strains, AB5075 and A118, were cultured overnight under different conditions (LB 423 broth and LB broth + 4% CSF). OD 600 was adjusted to 0.01 using M9 minimal media 424 and used for growth curves with different amino acids at 10 mM (Glutamine and Proline) 425 and carbon sources at 5 mg/mL (Glucose and Citrate). A microplate reader was used to 426 incubate the cultures at 37°C with shaking and OD 600 readings were taken every 15 427 minutes for 24 hours. Growth rate data was then analyzed using Prism 8 software. 428

Human Serum Growth Curves 429
Overnight cultures of A118 and AB5075 cultured in LB broth and LB broth + 4% CSF 430 were OD 600 adjusted (0.001) in in M9 minimal media and incubated along with either 431 10% human sera or 10% heat inactive human sera. Human sera were inactivated by 432 placing human sera in a 56℃ water bath for 30 minutes. Cultures were incubated at 37℃ 433 with shaking and OD 600 readings were taken every 15 minutes for 24 hours. Growth rate 434 data was analyzed using Prism 8 software. 435

Cytotoxicity assays. 436
In a Nunclon TM Delta Surface opaque 96-well microplate (ThermoScientific), we added 437 colorless DMEM, 4% CSF, and A118 or AB5075 CFCM diluted in LB broth to make 50 438 μL of CFCM at final concentrations of 1.5%, 5%, and 50%. An additional 50 μL of ATCC 439 HEK-293 cells at a concentration of 1 x 10 6 cells/mL in colorless DMEM were 440 suspended in the well and intoxicated for 1 h at 37°C, 5% CO 2 . CellTiter-Glo ® Reagent 441 (100 μl) was added to each experimental and standard curve well and then placed on 442 an orbital shaker for 2 min. Following mixing, plates were incubated at room 443 temperature for 10 min to stabilize the luminescent signal. The viability of HEK cells was 444 measured at room temperature using the "all" luminescence function of SpectraMax M3. 445

Galleria mellonella infection model 446
To assess the virulence of A. baumannii with and without CSF in vivo, the G. mellonella was used to assess overall larval viability for the duration of the assay. One hundred G. 455 mellonella larvae were used in each condition and incubated at (37 °C) in a sterile Petri 456 dish for 24 h intervals for 48 h total. Larvae viability was monitored by observing 457 response to gentle prodding with a glass rod; those with no response were considered 458 dead. Four replicates with 100 larvae per Petri dish were performed for each condition. 459

Drosophila melanogaster model of infection 460
The D. melanogaster population used in this study, population B1, is a large 461 outbred population, maintained on 14-day discrete generation cycles, 24:0 light:dark, 462 (or PBS control) and pricked into the ventrolateral thorax of each fruit fly, following the 470 pricking protocol described in Khalil et al. 2015 (58). After inoculation, flies were 471 maintained in groups of ten same-sex flies per vial on banana-molasses medium. Three 472 hours after inoculation, flies that died from handling were discarded, which was always 473 <5% of the flies. Thereafter, flies were transferred to fresh diet every three days. 474 Survival was recorded daily. 50 male flies were inoculated per treatment per each of 475 three biological replicates. 476

PNAG assays. 477
To study extracellular matrix (ECM) production, microcolony biofilm was used as model 478 The transcript levels of each sample were normalized to the recA rRNA transcript levels 498 for each cDNA sample. The relative quantification of gene expression was performed 499 using the comparative threshold method 2 -ΔΔCt . The ratios obtained after normalization 500 were expressed as folds of change compared with cDNA samples isolated from bacteria 501 cultures on LB. Statistical analysis (Mann-Whitney test) was performed using 502 GraphPad Prism (GraphPad software, San Diego, CA, United States). A P-value < 0.05 503 was considered significant. 504

ATP assay. 505
Both strains, A118 and AB5075, were cultured overnight in LB broth with or without 4% 506 CSF. Cultures were then diluted 1:100 using fresh LB broth and were incubated at 37℃ 507 with agitation. After 6 hours of incubation, an aliquot of each sample was taken. The 508 OD600 of each sample was recorded. Samples were prepared for the ATP assay using 509 the perchloric acid extraction method as previously described. 100 ul of ice-cold 1.2 M 510 perchloric acid was added to 200 ul of the bacterial sample and they were vortexed for 511 10 seconds. The mixture was then incubated on ice for 15 minutes. After incubation, the 512 samples were spin down at 16,100 x g for 5 minutes at 4℃. 200 ul of the supernatant 513 was then transferred into a fresh tube and mixed with a neutralizing solution (0.72 M 514 KOH and 0.16 M KHCO 3 ). The neutralized extract was spin down and the supernatant 515 was then transferred into a fresh tube and used for the ATP assay. 516 To measure intracellular ATP content, we followed the manufacturer instructions 517 (Promega, Madison, WI, United States). Briefly, 100 ul of the supernatant was added 518 into an opaque 96 well plate and allowed to stabilize to room temperature. 100 ul of the 519 Bac-Titer™ Glo Reagent was added into each of the samples and the plate was mixed 520 in an orbital shaker and incubated for 5 minutes. A standard curve was included in each 521 plate. Luminescence of the samples was recorded.    Table S1. Differential gene expression analysis of strain A118 under exposure to 758 4% CSF. RNA-seq read counts of 2 biological replicates of LB or LB with 4% CSF 759 treated A baumannii A118 were analyzed using DEseq software. For each hit, gene ID, 760 average base mean, base mean group A (LB-treated), base mean group B (LB+CSF 761 treated), fold-change, log 2 fold change, P-value, Benjamini-Hochberg adjusted P-value, 762 and gene description/function are provided. 763